#399

backend/app/controllers/post_imports_controller.rb

@@ -61,6 +61,13 @@ class PostImportsController < ApplicationController
     columns = Array(value['columns'])
     rows = Array(value['rows'])
     raise ArgumentError, '解析結果の形式が不正です．' unless columns.all?(Hash) && rows.all?(Hash)
+    raise ArgumentError, '取込件数が多すぎます．' if rows.length > PostImportSourceParser::MAX_ROWS
+    raise ArgumentError, '列数が多すぎます．' if columns.length > 50
+    unless rows.all? { |row| Array(row['values']).length <= columns.length &&
+                               Array(row['values']).all? { _1.is_a?(String) && _1.bytesize <= PostImportSourceParser::MAX_CELL_BYTES } }
+      raise ArgumentError, '解析結果のセルが不正です．'
+    end
+    raise ArgumentError, '解析結果が大きすぎます．' if rows.sum { Array(_1['values']).sum(&:bytesize) } > PostImportSourceParser::MAX_BYTES
 
     { columns: columns.map { |column| column.slice('key', 'label') },
       rows: rows.map { |row| { source_row: row['source_row'], values: Array(row['values']) } } }

backend/app/services/post_import_previewer.rb

@@ -26,6 +26,12 @@ class PostImportPreviewer
       should_fetch = fetch_metadata == true || fetch_metadata.to_i == row[:source_row].to_i
       metadata, fetch_warnings = metadata_for(url, should_fetch, metadata_cache, fetch_warnings)
       metadata.each do |field, value|
+        if field == 'tags' && provenance[field] != 'manual'
+          previous_origin = provenance[field]
+          attributes[field] = [attributes[field], value].compact.join(' ').split.uniq.join(' ')
+          provenance[field] = previous_origin == 'automatic' ? 'automatic' : previous_origin
+          next
+        end
         next unless provenance[field] == 'automatic' || attributes[field].blank?
 
         attributes[field] = value

backend/app/services/post_metadata_fetcher.rb

@@ -7,21 +7,39 @@ class PostMetadataFetcher
     content = -> name { document.at_css("meta[property='#{ name }'], meta[name='#{ name }']")&.[]('content')&.strip.presence }
     duration = content.call('og:video:duration') || content.call('video:duration')
     published = content.call('article:published_time') || content.call('date')
-    published_at = Time.zone.parse(published.to_s) if published.present?
+    created_range = original_created_range(published)
     platform_tags = platform_tags(uri)
     { title: metadata[:title],
       thumbnail_base: Preview::KnownSiteExtractor.thumbnail_url(uri) || metadata[:image_url],
-      original_created_from: published_at&.iso8601,
+      original_created_from: created_range&.first&.iso8601,
-      original_created_before: published_at && (published_at + 1.minute).iso8601,
+      original_created_before: created_range&.last&.iso8601,
       duration: duration&.to_f&.then { _1.positive? ? (_1 * 1_000).round : nil },
       tags: platform_tags.join(' ') }
   end
 
   def self.platform_tags uri
-    return ['動画', 'YouTube'] if Preview::KnownSiteExtractor.thumbnail_url(uri)
+    return ['動画', 'YouTube'] if Preview::KnownSiteExtractor.youtube_video_id(uri)
     return ['動画', 'ニコニコ'] if Preview::KnownSiteExtractor.niconico_video_id(uri)
 
     []
   end
-  private_class_method :platform_tags
+
+  def self.original_created_range value
+    return nil if value.blank?
+
+    raw = value.to_s.strip
+    from = Time.zone.parse(raw)
+    before =
+        case raw
+        when /\A\d{4}\z/ then from + 1.year
+        when /\A\d{4}-\d{2}\z/ then from + 1.month
+        when /\A\d{4}-\d{2}-\d{2}\z/ then from + 1.day
+        when /:\d{2}(?:Z|[+-]\d{2}:?\d{2})?\z/ then from + 1.second
+        else from + 1.minute
+        end
+    [from, before]
+  rescue ArgumentError, TypeError
+    nil
+  end
+  private_class_method :platform_tags, :original_created_range
 end

backend/app/services/preview/known_site_extractor.rb

@@ -4,6 +4,15 @@ module Preview
       youtube_thumbnail(uri)
     end
 
+    def self.youtube_video_id(uri)
+      case uri.host&.downcase
+      when 'youtu.be'
+        uri.path.split('/').reject(&:blank?).first
+      when 'www.youtube.com', 'youtube.com', 'm.youtube.com'
+        uri.path == '/watch' ? URI.decode_www_form(uri.query.to_s).to_h['v'] : nil
+      end&.then { _1 if _1.match?(/\A[A-Za-z0-9_-]{6,20}\z/) }
+    end
+
     def self.niconico_video_id(uri)
       case uri.host&.downcase
       when 'www.nicovideo.jp', 'nicovideo.jp'
@@ -14,14 +23,8 @@ module Preview
     end
 
     def self.youtube_thumbnail(uri)
-      id =
+      id = youtube_video_id(uri)
-          case uri.host&.downcase
+      return unless id
-          when 'youtu.be'
-            uri.path.split('/').reject(&:blank?).first
-          when 'www.youtube.com', 'youtube.com', 'm.youtube.com'
-            uri.path == '/watch' ? URI.decode_www_form(uri.query.to_s).to_h['v'] : nil
-          end
-      return unless id&.match?(/\A[A-Za-z0-9_-]{6,20}\z/)
 
       "https://i.ytimg.com/vi/#{ id }/hqdefault.jpg"
     end

frontend/src/pages/posts/PostImportPage.tsx

@@ -30,6 +30,7 @@ type ImportRow = {
     status: 'ready' | 'warning' | 'error'
     existing: boolean }
 type ParsedImport = { columns: Column[], rows: { sourceRow: number, values: string[] }[] }
+type Phase = 'source' | 'mapping' | 'preview' | 'result'
 
 
 const detectFormat = (source: string): 'csv' | 'tsv' =>
@@ -42,6 +43,7 @@ const publicGoogleSheetURL = (source: string): boolean =>
 
 const PostImportPage: FC<Props> = ({ user }) => {
   const [source, setSource] = useState ('')
+  const [phase, setPhase] = useState<Phase> ('source')
   const [format, setFormat] = useState<'csv' | 'tsv' | 'json' | 'google_sheets'> ('tsv')
   const [hasHeader, setHasHeader] = useState (false)
   const [jsonPath, setJsonPath] = useState ('')
@@ -69,6 +71,15 @@ const PostImportPage: FC<Props> = ({ user }) => {
       setParsed (data)
       if (data.columns.length === 1)
         setUrlColumn ('0')
+      else
+        {
+          const candidate = data.columns.map ((_, index) => ({ index,
+              count: data.rows.filter (row => /^https?:\/\//.test (row.values[index] ?? '')).length }))
+                                      .sort ((a, b) => b.count - a.count)[0]
+          if (candidate?.count)
+            setUrlColumn (String (candidate.index))
+        }
+      setPhase ('mapping')
     }
     catch
     {
@@ -92,6 +103,7 @@ const PostImportPage: FC<Props> = ({ user }) => {
         mappings,
         existing })
       setRows (data.rows)
+      setPhase ('preview')
     }
     catch
     {
@@ -168,6 +180,7 @@ const PostImportPage: FC<Props> = ({ user }) => {
                                                                                               { rows: targets })
       toast ({ title: `登録完了: ${result.created} 件`,
               description: `スキップ ${result.skipped} 件、失敗 ${result.failed} 件` })
+      setPhase ('result')
     }
     catch
     {
@@ -186,7 +199,7 @@ const PostImportPage: FC<Props> = ({ user }) => {
       <MainArea>
 	<Helmet><title>{`投稿インポート | ${SITE_TITLE}`}</title></Helmet>
 	<PageTitle>投稿インポート</PageTitle>
-	{rows.length === 0 ? (
+	{phase === 'source' ? (
 	    <Form>
 	      <p className="text-sm text-neutral-600 dark:text-neutral-300">
 		URL を一行に一件ずつ貼り付けるか、CSV・TSV・JSON を入力してください。
@@ -224,7 +237,7 @@ const PostImportPage: FC<Props> = ({ user }) => {
 		非公開シートには対応していません。ウェブ公開 URL を使用するか、CSV を保存して選択、又は内容を貼り付けてください。
 	      </p>}
 	      <Button type="button" onClick={parseSource} disabled={loading || !(source)}>列を解析</Button>
-	    </Form>) : parsed && rows.length === 0 ? (
+	    </Form>) : phase === 'mapping' && parsed ? (
 	    <Form>
 	      <PageTitle>列の割当て</PageTitle>
 	      <p>URL 列と補助列を指定してから、投稿情報を取得します。</p>
@@ -238,8 +251,10 @@ const PostImportPage: FC<Props> = ({ user }) => {
 		  <option value="">指定なし</option>
 		  {columns.map ((column, index) => <option key={column.key} value={index}>{column.label}</option>)}
 		</select></label>)}
+	      <div className="overflow-x-auto"><table className="text-sm"><thead><tr>{columns.map (column => <th key={column.key}>{column.label}</th>)}</tr></thead><tbody>{parsed.rows.slice (0, 5).map (row => <tr key={row.sourceRow}>{row.values.map ((value, index) => <td key={index}>{value}</td>)}</tr>)}</tbody></table></div>
+	      <Button type="button" variant="outline" onClick={() => setPhase ('source')}>入力元へ戻る</Button>
 	      <Button type="button" onClick={preview} disabled={loading || urlColumn === ''}>投稿情報を取得して確認</Button>
-	    </Form>) : (
+	    </Form>) : phase === 'preview' ? (
 	    <div className="space-y-4">
 	      <div className="flex flex-wrap gap-3 rounded border p-3 dark:border-neutral-700">
 		<label>URL 列 <select value={urlColumn} onChange={ev => setUrlColumn (ev.target.value)}>
@@ -258,7 +273,12 @@ const PostImportPage: FC<Props> = ({ user }) => {
 	      <div className="space-y-3 md:hidden">{rows.map ((row, index) => <RowCard key={row.sourceRow} row={row} index={index} update={updateAttribute}/>)}</div>
 	      <div className="hidden overflow-x-auto md:block"><table className="w-full text-sm"><thead><tr><th>行</th><th>URL</th><th>タイトル</th><th>サムネール</th><th>日時</th><th>動画時間</th><th>タグ</th><th>親投稿</th><th>状態</th></tr></thead>
 		<tbody>{rows.map ((row, index) => <tr key={row.sourceRow} className="border-t dark:border-neutral-700"><td>{row.sourceRow}</td><td><input value={row.url} onChange={ev => updateURL (index, ev.target.value)}/></td><td><input value={row.attributes.title ?? ''} onChange={ev => updateAttribute (index, 'title', ev.target.value)}/></td><td><input value={row.attributes.thumbnailBase ?? ''} onChange={ev => updateAttribute (index, 'thumbnailBase', ev.target.value)}/></td><td><input value={row.attributes.originalCreatedFrom ?? ''} onChange={ev => updateAttribute (index, 'originalCreatedFrom', ev.target.value)}/><input value={row.attributes.originalCreatedBefore ?? ''} onChange={ev => updateAttribute (index, 'originalCreatedBefore', ev.target.value)}/></td><td><input value={String (row.attributes.duration ?? '')} onChange={ev => updateAttribute (index, 'duration', ev.target.value)}/></td><td><input value={row.attributes.tags ?? ''} onChange={ev => updateAttribute (index, 'tags', ev.target.value)}/></td><td><input value={row.attributes.parentPostIds ?? ''} onChange={ev => updateAttribute (index, 'parentPostIds', ev.target.value)}/></td><td>{row.status}<FieldError messages={[...row.warnings, ...Object.values (row.errors).flat ()]}/></td></tr>)}</tbody></table></div>
+	      <Button type="button" variant="outline" onClick={() => setPhase ('mapping')}>列割当てへ戻る</Button>
 	      <Button type="button" onClick={submit} disabled={loading}>有効な投稿を一括登録</Button>
+	    </div>) : (
+	    <div className="space-y-3"><p>インポート処理が完了しました。</p>
+	      <Button type="button" onClick={() => setPhase ('preview')}>結果を確認</Button>
+	      <Button type="button" variant="outline" onClick={() => setPhase ('source')}>新しい入力元を選ぶ</Button>
 	    </div>)}
 	</MainArea>)
 }